$NBIS is increasingly looking like much more than another company benefiting from AI infrastructure demand. The real opportunity is in the platform being built around that demand. $NBIS is not just another customer buying accelerators. $NVDA has invested directly into the company, the two are working closely on AI factory architecture and software, and $NBIS is already preparing infrastructure around the latest generations of hardware, including Vera Rubin. That matters because the next phase of AI infrastructure will not be won simply by whoever owns the most GPUs. The advantage will belong to platforms that can deploy them efficiently, connect them with the right networking and software stack, and make that infrastructure easy for customers to use at scale. That is exactly where $NBIS is positioning itself. Token Factory takes the company deeper into inference, model serving and fine-tuning. Eigen AI adds another layer of optimization. Instead of simply selling access to compute, $NBIS is building further up the stack and becoming more involved in how models are deployed, optimized and run in production. That could become increasingly important as inference grows into a much larger part of AI spending. Then there is Physical AI. Robotics, autonomous systems and synthetic-data workloads are enormously compute-intensive, and $NBIS is already building infrastructure for that market together with $NVDA. If robotics becomes one of the next major AI compute cycles, $NBIS will already have the infrastructure and relationships in place to participate. The partnership with $PLTR adds another angle. Sovereign AI is becoming increasingly important as enterprises and governments care more about where models run, where data is stored and who controls the underlying infrastructure. That gives $NBIS exposure to a very different market from traditional cloud demand. At the same time, $MSFT and $META provide another level of validation. These companies do not need basic GPU capacity. They need infrastructure that can operate reliably at enormous scale and meet demanding technical requirements. This is what makes the broader $NBIS opportunity so compelling. Next-generation $NVDA hardware, hyperscale AI infrastructure, inference, model optimization, sovereign AI, enterprise workloads and eventually Physical AI are increasingly converging on the same platform.

$ALMU approach to integrating high-performance semiconductor materials could become increasingly relevant as AI infrastructure demands faster optical connectivity. My latest article examines the technology, its commercial potential and the milestones that matter for shareholders. The investment case deserves a closer look. Full article ↓

I think $NVTS and $INFQ are both getting close to points where the story can change materially. For $NVTS, this is becoming much more than a GaN story. The $NVDA 800V HVDC collaboration gives the company exposure to one of the biggest challenges in AI infrastructure: getting power into increasingly dense compute systems efficiently. Then you add SiC, Claros and the $MX partnership. Piece by piece, $NVTS is building a much broader position across the power stack instead of depending on one product category. As AI infrastructure reaches higher rack densities and megawatt-scale systems, power becomes a much bigger part of the investment case. That is where $NVTS has a real opportunity. $INFQ is different, but the setup is just as interesting. The company is starting to show that it can go beyond pure quantum research and into real commercial use cases. Revenue is growing, government programs are getting larger, and the progress on logical qubits is becoming more tangible. NASA, Eaton and $CSCO also give the story more credibility because they show that the technology is being tested and used outside the lab.

$SIVE The US listing story is getting harder and harder to ignore. What started only a few months ago as an evaluation of a potential Nasdaq dual listing is increasingly looking like a real corporate project with a defined timeline, dedicated infrastructure and very concrete preparation behind it. Yesterday’s update is another important step. Sivers is proposing Ernst & Young as its new auditor and explicitly connects that decision to its preparations for a potential US dual listing, US capital markets experience and SEC reporting requirements. That matters. The company is actively changing parts of its financial and reporting infrastructure to make itself compatible with the requirements of the US capital markets. Look at how quickly the language has evolved. In April, $SIVE said it was evaluating a potential dual listing. Then the company started preparing historical financial statements under PCAOB requirements. In July, management referred to its dual listing objective. In August, $SIVE said it expected to complete all necessary preparations during the first half of 2027. Now we have a proposed auditor change where US capital markets experience and SEC reporting capability are explicitly cited as reasons. That progression is extremely important. Evaluation. PCAOB preparation. Internal reporting upgrades. A defined H1 2027 preparation timeline. A new auditor with US and SEC experience. Its photonics opportunity is expanding rapidly. The company is building toward significantly larger CW laser capacity in Glasgow. Its relationship with $JBL creates exposure to the next generation of AI data center optical interconnects. Its technology is positioned around external laser sources, silicon photonics, optical I/O and the broader shift toward higher bandwidth and lower power connectivity inside AI infrastructure. At the same time, $SIVE continues to build exposure to US defense, satellite communications and advanced semiconductor programs. That means a US listing would not simply be a cosmetic change of trading venue. It could place $SIVE directly in front of the investor base that already understands and actively values AI infrastructure, photonics, optical connectivity, semiconductors, defense technology and next-generation data center architectures. That investor base is dramatically deeper in the United States than it is in Sweden. And this is what makes the potential timing so fascinating. $SIVE currently expects the necessary US listing preparations to be completed during H1 2027. The major commercial ramp in photonics is expected to become increasingly visible through 2027. Glasgow capacity expansion is targeted for late 2027. The company is positioning itself for a much larger revenue base beyond that. In other words, $SIVE could potentially arrive on the US market just as the fundamental story becomes substantially easier for investors to see. For a small Swedish semiconductor company operating directly inside some of the most important technology transitions of this decade, that matters enormously. Companies do not prepare PCAOB financials, upgrade reporting systems, establish a H1 2027 preparation timeline and select an auditor specifically for US capital markets and SEC requirements for no reason. For me, the US listing is rapidly developing from an interesting optionality into one of the most important potential catalysts in the entire $SIVE story.

$SIVE $INFQ $FLY $CCXI $WULF You can choose only ONE stock from each sector 👇 Photonics Quantum Space Robotics AI Infrastructure 5 stocks. 5 sectors. No changes until 2030. Which names make your portfolio? Drop your 5 picks below - and repost this so we can see which companies the community has the highest conviction in.

$LITE $AAOI $SIVE $CRDO The more I look at the next phase of data center expansion, the more convinced I am that optical connectivity will become one of the most important parts of the entire infrastructure buildout. GPU performance keeps increasing, clusters keep getting larger and bandwidth requirements are rising at an extraordinary rate. At some point, adding more compute is not enough. The data still has to travel between GPUs, racks and data centers quickly enough, while power consumption and latency remain under control. That is exactly where these four companies become interesting. $LITE sits at one of the most critical points of the optical supply chain. Its expertise in InP lasers and high power laser technology gives the company exposure to several architectures that could become increasingly important as the industry progresses from traditional pluggable optics toward external laser sources, NPO and eventually CPO. At the same time, demand for 800G and 1.6T continues to increase, creating a very strong environment for optical component suppliers with meaningful manufacturing capacity. $AAOI is becoming increasingly interesting because the company is no longer just positioned around a future opportunity. It is already seeing significant demand for 800G and 1.6T products and is expanding manufacturing capacity to support hyperscaler deployments. If the transition toward 1.6T accelerates as expected, $AAOI has the potential to become a much more important supplier within the data center optical ecosystem. $SIVE is obviously the smallest company of the group, but that is also what makes the setup interesting. Sivers is building its position around InP DFB lasers, one of the components required to generate the optical light used inside these high speed systems. The $JBL 1.6T LRO program is particularly important because successful qualification and production orders would take $SIVE from development exposure toward real volume production. On top of that, the company is expanding its own InP capacity while also maintaining access to external foundries, giving it more flexibility if demand develops faster than expected. Then there is $CRDO, which has gradually become much more than an AEC story. The company is building a broader connectivity platform around SerDes, DSP technology, optical solutions and silicon photonics. That gives Credo exposure to several different ways hyperscalers can solve the same fundamental problem, which is moving enormous amounts of data through increasingly dense compute infrastructure. What makes this group so interesting is that they are not competing for exactly the same piece of the market. $LITE provides critical optical and laser technology. $AAOI is scaling high speed optical transceivers. $SIVE provides the InP laser technology that can sit deeper inside the optical supply chain. $CRDO is building the silicon and connectivity layer around the entire system. Different positions in the stack, but all benefiting from the same structural pressure. Compute is scaling extremely fast. Networking now has to catch up. And as clusters become larger, faster and more power hungry, the value of the technology connecting those clusters could become just as important as the processors themselves.

Something worth paying attention to at $MX. CEO Chae Lee just bought 100,000 shares of $MX with his own money, investing roughly $318K across two days. The timing is what makes this interesting. Just days earlier, $NVTS agreed to invest $5M into $MX at $3.42 per share as the two companies deepen their partnership around high-voltage SiC power semiconductors. Strategic investment from $NVTS. Significant insider buying from the $MX CEO. An expanding SiC partnership on top of an established MOSFET business. Management and its strategic partner are both putting real capital behind the story.

$NVDA $TSLA $AAPL $AMD I’ve built Crash Capital, a free Wall Street browser game for the investing community. Choose your character: Pep @Pep_Invest Mike @EhrmantrautCap_ Elon @elonmusk Jensen @JensenHuang You have 90 seconds to collect Shares, Gold, Silver and Bitcoin, dodge the bears and bank your gains before the closing bell. Only the money you safely deposit counts toward your final score. Play here: Give it a try and let me know what you think. What works well, and what would you change or add? Once you’ve played, post a screenshot of your final score in the comments. Let’s see who can break $100,000 👇

$NVDA $TSLA $AAPL $AMD I’ve built Crash Capital, a free Wall Street browser game for the investing community. Choose your character: Pep @Pep_Invest Mike @EhrmantrautCap_ Elon @elonmusk Jensen @JensenHuang You have 90 seconds to collect Shares, Gold, Silver and Bitcoin, dodge the bears and bank your gains before the closing bell. Only the money you safely deposit counts toward your final score. Play here: Give it a try and let me know what you think. What works well, and what would you change or add? Once you’ve played, post a screenshot of your final score in the comments. Let’s see who can break $100,000 👇

$SIVE already has a commercial relationship with Doosan. The interesting question is whether it could eventually extend beyond satellite communications. Sivers signed a $1.5 million development agreement with Doosan Corporation for Ka-band satellite antenna systems, contributing its beamforming technology and antenna design while Doosan handles manufacturing and system testing. The agreement was described as the beginning of a broader SATCOM partnership. Separately, Doosan Tesna, the semiconductor testing business within the same group, has historically served Samsung Electronics and SK hynix. Tesna has also been suggested as the sole-source wafer-test supplier for $NVDA Groq 3, although that claim remains unconfirmed. If confirmed, it would place another Doosan business directly in that production supply chain. What interests me is whether $SIVE existing engineering relationship with Doosan could eventually support introductions, technical discussions or commercial opportunities elsewhere within the group’s semiconductor network. $SIVE also develops lasers for optical connectivity in AI data centers, so a future collaboration involving photonics would be particularly relevant. That is speculation, not an announced expansion of the partnership. Doosan Corporation’s Electro-Materials business and Doosan Tesna are different operations. But $SIVE already has a paid development agreement with Doosan, rather than simply operating in an adjacent market. That makes the future scope of the relationship worth following. The development I would watch for is a separate photonics collaboration, product qualification or supply agreement. Until then, the confirmed business is SATCOM. The potential opportunity is whether that relationship can grow beyond it.

I gave my subscribers $MX; the stock is set for the next bull run - the fundamentals are just too good. When will you join my inner circle for $3 a month?

Who are your top 10 accounts with fewer than 20,000 followers? @FinnStockinger @EhrmantrautCap_ @HunterAllen4 @corleoneinvests @Sofigoodboy @Million_Sancet @awodias @outliercapx @athuinvests @CaesarCapitalz Tag them below and repost this. Let’s give smaller accounts putting real work into their research and analysis the attention they deserve.

It is incredible that, despite having 15,000 followers, strong original content, and numerous well-researched articles, one still falls below the minimum earnings threshold; on one hand, an immense amount of work goes into it, and currently, one is dependent on subscribers. On the other hand, it motivates me to improve even further. I am in the process of creating something big - a research platform offering a level of depth and information density never seen before. It will primarily feature stocks that I own myself, have on my watchlist, or that are frequently discussed on X. If you have any ideas regarding what is particularly important to you for the platform, please feel free to leave a suggestion.

$SIVE could become a substantially larger business as AI infrastructure adopts optical connectivity. What keeps me bullish is the combination of its laser technology, the companies developing products around it, and the manufacturing capacity now being planned. $SIVE Indium phosphide lasers provide the light used in optical data connections. Its portfolio includes 70 mW and 100 mW devices for 800G and 1.6T transceivers, alongside laser arrays for co-packaged optics and optical I/O. That gives the company exposure to several stages of the industry’s development. The detail I think deserves considerably more attention is how these devices are manufactured. $SIVE process allows full optical testing while the devices are still on the wafer, before they are cut into individual chips. Its laser arrays also reduce the alignment steps required during assembly. The company identifies both as ways to improve manufacturing efficiency and reduce costs. At millions of units, those details become commercially significant. Identifying defective devices earlier and simplifying assembly can improve the economics of supplying an entire product generation. This is where I see an important part of the technology’s potential: performance that is designed to be manufactured repeatedly, efficiently and at scale. The planned Glasgow expansion makes that ambition concrete. $SIVE has announced a $30 million investment targeting capacity for more than 100 million CW DFB lasers annually, with the expanded facility expected to become operational in Q4 2027. That capacity would complement its external manufacturing partners, giving the company several routes to support growing customer volumes. The commercial relationships help explain why management is preparing for that scale. $JBL is developing a 1.6T LRO transceiver using $SIVE lasers. @Ayar has demonstrated $SIVE arrays in its SuperNova light source. $GFS has announced plans to incorporate $SIVE arrays into silicon photonics reference designs and make them available through its SCALE platform. Together, these relationships create potential routes into pluggable optics, optical I/O and co-packaged systems. The $GFS relationship is especially interesting to me. Inclusion in reference designs could make $SIVE technology accessible to multiple companies developing different products. Each successful adoption could extend the commercial return on the underlying technology. That is where the scaling potential becomes compelling: a common technology platform supporting multiple customers, followed by repeat shipments as their products enter production. $SIVE is already reallocating resources toward that transition, prioritizing upcoming product ramps over development work. Management expects the shift to accelerate through 2027. If successful, it could allow revenue to grow substantially faster than the engineering effort required for each additional sale. $SIVE has a credible opportunity to commercialize the same underlying expertise across several large optical markets.

I gave my subscribers $MX; the stock is set for the next bull run - the fundamentals are just too good. When will you join my inner circle for $3 a month?

$IBM selected $MRAM for FlashCore. Amentum is behind its $40M defense agreement. $MXL is evaluating $MRAM for AI servers. I dug into @EverspinTech customers, partners, growth opportunities and risks. @PolarizingLit @HunterAllen4